- Email: [email protected]
Chemical and cellulose crystallite changes in Pinus radiata during torrefaction
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
Available online at www.sciencedirect.com
http://www.elsevier.com/locate/biombioe
Chemical and cellulose crystallite changes in Pinus radiata during torrefaction Stefan J. Hill*, Warren J. Grigsby, Peter W. Hall Scion, Private Bag 3020, Rotorua 3046, New Zealand
article info
abstract
Article history:
The impact on the chemical composition and changes to the cellulose crystallites in Pinus
Received 1 July 2012
radiata wood chips under light (ca. 230 C), mild (ca. 260 C), and severe (290þ C) torrefaction
Received in revised form
temperatures at a range of times was examined by solid state
25 April 2013
Dephased NMR spectroscopy, TGA, and synchrotron based X-ray diffraction. Results
Accepted 26 April 2013
indicated the decomposition of hemicelluloses into furfurals at the lowest temperature
Available online 30 May 2013
with little modification to the lignin or cellulose. De-polymerisation of lignin and cellulose
13
C CP-MAS and Dipolar
was observed as torrefaction severity increased. The increased hydrophobicity under light Keywords:
and mild torrefaction severity was attributed to a combination of thermo-chemical mod-
Moisture content
ifications to hemicelluloses and lignin, along with cellulose crystal lattice changes. The
Torrefaction
observed decrease in hydrophobicity under severe torrefaction conditions was attributed
Softwood
to the degradation of cellulose crystallites.
Solid state NMR
ª 2013 Elsevier Ltd. All rights reserved.
X-ray diffraction
1.
Introduction
Torrefaction is the semi-conversion of wood to charcoal by pyrolysing lignocellulosics at temperatures ranging from 200 to 300 C in an inert or semi-inert atmosphere at standard pressures [1]. This has the effect of improving not only the energy density [2], but also moisture stability [3,4] and grindability [5] of the woody biomass. This is beneficial when considering transport costs, long term storage of torrefied wood [6], and processing into pulverised fuel. The torrefaction process has been shown to cause many changes to the chemical composition of woody biomass depending on the temperature and duration of heating [7]. These changes include modifications to the lignin and hemicelluloses at lower torrefaction temperature and degradation of cellulose, lignin, and the hemicelluloses at higher torrefaction temperatures.
Previous studies have illuminated some of the chemical processes occurring during torrefaction utilising techniques such as NMR, FT-IR, and X-ray diffraction to name a few [1,3]. Chen and Kuo [8] defined three temperature regimes of torrefaction; light torrefaction at temperatures ca. 230 C, mild torrefaction at temperatures ca. 260 C, and severe torrefaction at temperatures of 290 C and above. These definitions can be described in terms of the thermal modification extent of the wood components during torrefaction. Under light conditions the mass fractionloss of the cellulose (1.1%), hemicelluloses (2.7%), and lignin (1.5%) is minimal. Under mild conditions and the mass fraction losses increase to 4.4%, 38.0%, and 3.1% respectively, meaning that at ca. 260 C there is proportionally significantly more degradation of the hemicelluloses than either cellulose or lignin. Under the severe torrefaction regime further mass fraction losses in the three components occur: 44.8% for cellulose, 58.3% for the hemicelluloses, and 7.0% for
* Corresponding author. Scion, 49 Sala St, Rotorua 3010, New Zealand. Tel.: þ64 7 343 5872; fax: þ64 7 343 5507. E-mail address: [email protected] (S.J. Hill). 0961-9534/$ e see front matter ª 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biombioe.2013.04.025
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
lignin. It should be noted that these substances were tested as pure compounds and not in wood [8]. These provide a useful guide to the influence of temperature on the thermal degradation of wood. It has been long known that the thermal modification of wood leads to an increase in hydrophobicity and that this is a consequence of changes to the nano-structure of the wood [9,10]. This observation has been attributed to coupling and decoupling of hydroxyl groups on the cellulose crystal surface [11]. The nature of these changes has more recently been associated in part to partially irreversible changes to the cellulose chain conformations on crystal surfaces [12]. In the current study two technical properties of Pinus radiata wood, notably hydrophobicity and grindability, were studied as potential functions of overall chemical and nanostructural changes. This was achieved by torrefying wood at several combinations of temperature and times. Specimens were taken for characterisation by NMR, X-ray diffraction, TGA, and elemental analysis.
2.
Materials & methods
2.1.
Sampling
The trees (Pinus radiata D. Don) used in this study were w28 years old, sourced from log production site 191100E 5742800N (NZTM) in Kaingaroa forest in May of 2010, on the volcanic plateau in the Central North Island of New Zealand (38 340 S, 176 320 E). The wood chip used in the experiments was recovered from chipped sawmilling slab wood, and so is from the outer more dense part of the stem, with the log being cut from the lower part (bottom 12 m) of the tree. Due to the industrial nature of the chipping process, the sample should be regarded as a bulked up average. Logs were processed from stump to mill in 3e4 days, with open air storage at harvest and sawmill sites.
2.2.
Torrefaction
A torrefaction test rig used a small electric furnace fitted with a torrefaction reactor tube diameter (54 mm diameter by 240 mm long; 0.55 L). The reactor tube was fed with argon gas which was preheated in a copper coil before entering the reactor tube. The temperature of the furnace, the internal temperature of the reactor tube (sides and centre), and the electrical power use of the furnace were monitored during torrefaction. Moisture contents (MC%) of samples were calculated as a percentage of asamples 105 C oven-dried weight: MCð%Þ ¼ ðOriginal massðgÞ=Oven dried massðgÞ 1Þ 100 (1) Torrefaction experiments used air dried (average moisture content 18%) P. radiata chips (9e26 mm in size), temperatures of 220 C, 260 C, and 300 C, and torrefaction times of 5, 30, and 60 min. In all cases the heating rate of the furnace was 8 C$min1 until the target temperature was reached. The
93
specimens were then held at the target temperature for the duration of the torrefaction experiment. After the prescribed time the rig was opened and the specimen was removed and allowed to air-cool. Sub-samples of these specimens were used for NMR, TGA, hydrophobicity, grindability, and elemental analysis experiments. A further two specimens were torrefied at 250 C and 275 C for 45e50 min, i.e. at temperatures identified as being in the range of particular interest, for use in synchrotron-sourced X-ray diffraction studies. These specimens were conditioned at 22 C and 32% relative humidity for at least 48 h prior to further analysis.
2.3.
NMR
The samples were spun at 5 kHz in a 4 mm Bruker SB magicangle spinning probe, for 13C NMR at 50.3 MHz using a Bruker 200 DRX spectrometer. For the standard cross-polarisation (CP) experiment, each 1.5 s pulse delay was followed by a proton preparation pulse of duration tp ¼ 4.6 ms, a 1 ms contact time and a 30 ms acquisition time. The proton transmitter power was increased to a value corresponding to a 90 pulse width of 2.8 ms for proton decoupled during 13C data acquisition. In case of dipolar dephasing experiments a dephasing delay of 70 ms was introduced on the carbon channel prior to decoupled data acquisition. Transients were averaged over 10k transients for CP and 60k transients for dipolar dephasing experiments. All spectra had a Gaussian line broadening of 25 Hz applied prior to Fourier transform and were calibrated so that the cellulose interior C4 peak was assigned a value of 89.3 ppm, previously established relative to polydimethylsilane at 1.96 ppm, in turn measured relative to tetramethylsilane at 0 ppm [13].
2.4.
X-ray diffraction
A synchrotron radiation sourced X-ray beam energy of 13 keV ˚ . A Mar-CCD was used, giving an X-ray wavelength of 0.9537 A detector (1024 1024 pixels) was located 120 mm from the specimen allowing detection of 2q angles out to approximately 35 . The beam size was 0.2 mm by 0.1 mm, and the detector pixel size was 0.158 mm2. The exposure time was 2 s with nine data points taken per specimen spaced 2e3 mm apart to obtain nine diffraction patterns per specimen. The diffractograms presented are the average of the nine data sets obtained per specimen. Diffraction traces were extracted from diffraction patterns using an approximate 10 cake slice orthogonal to the grain direction. From this an approximate 10 cake slice centred approximately 30 off-axis from the grain direction was subtracted. This subtraction removes the background from the diffuse halo of reflections of non-cellulose origin such as hemicelluloses and water [12]. A Lorentzian function was used to fit the major peaks assuming, for simplicity, a totally cellulose Ib structure (d(10), d(1 1 0), d(200)) [14]. Although P. radiata consists of a mixture of cellulose Ia and Ib crystal forms of almost equal proportions [15], with peaks that overlap significantly, the major structural difference seen is in the fibre repeat direction and as such is not significant when calculating cross-sectional dimensions [16].
94
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
The minimum lateral crystal dimension (L) was calculated using the Scherrer equation: Lhkl ¼ Kl=ßhkl cosqhkl
(2)
where K is a crystal geometry parameter, approximated as 0.93 [17], l is a x-ray wavelength, b is the width at half peak height, and q is the Bragg angle. As the value of L is dependent on the crystal plane chosen, Lð110Þ , L(1 1 0), and L(2 0 0) were calculated, and averaged to give an Laverage value. An estimate of the proportion of surface to interior cellulose crystal chains was made using [15]: 2 X ¼ Laverage 2h =L2average
(3)
where L is the calculated minimum lateral crystal dimension and h is the thickness of a cellulose chain (0.57 nm). An estimate of changes in crystallinity was determined by the empirical method of Segal [18]: C:I: ¼ Ið2 0 0Þ Iam =Ið2 0 0Þ
(4)
where I(2 0 0) is the height of the (2 0 0) peak, and Iam is the height of the minima between the d(1 1 0) and the d(2 0 0) peaks.
2.5.
TGA
Thermogravimetric analysis (TGA) was undertaken to simulate the torrefaction process using a Thermal Analysis Instruments Q500 instrument under a nitrogen atmosphere. Air dried wood specimens were heated at a rate of 50 C$min1 until the required isothermal temperature (220, 260, or 300 C) was reached. The specimens were held at this isothermal temperature for a further 50 min with weight loss recorded.
2.6.
Other analysis
Elemental analysis (%C/%H) was carried out by The Campbell Microanalytical Laboratory (University of Otago, New Zealand), %O was estimated by difference. Grindability was determined by grinding samples in a ring-mill for 1 min with the resulting material sieved to determine particle size distributions (>500 mm, 100e500 mm, and <100 mm).
3.
Results & discussion
Mass loss of P. radiata chips during torrefaction was measured from TGA experiments, with mass losses after 10 min of 7.3%, 14.8%, and 36.6% for 220 C, 260 C, and 300 C respectively. After 60 min this loss increased to 8.9%, 27.7%, and 68.2% for 220 C, 260 C, and 300 C respectively (Fig. 1). These weight loss figures are consistent with those previously reported for willow at similar temperatures and times where the mass loss was suggested to be a two-step kinetic process, with the first being the decomposition of hemicelluloses (Ea 76.0 kJ$mol1) and the second the decomposition of cellulose (Ea 151.7 kJ$mol1) [19]. A study by Chen and Kuo of the mass loss of individual lignocellulosic components (cellulose, hemicelluloses, and lignin) agrees with this two-step kinetic description, with lignin loss being only a relatively
Fig. 1 e TGA analysis of mass loss of wood at 220 C (line), 260 C (dash), and 300 C (dot-dash).
minor contributor to mass loss at these temperatures studied [8]. Mass loss during actual torrefaction was found to be 3.2%, 17.0%, and 42.6% for 220, 260, and 300 C respectively, all after 60 min. It can be seen in all cases that the TGA mass losses overestimate the amount of mass loss observed in the torrefaction system used. The difference is attributed to a combination of increased heating ramp and larger surface area of the small samples required for TGA analysis. Increased time and temperature of torrefaction was associated with an increase in grindability and resulted in greater fraction of sub-100 mm particle sizes (Table 1).
Table 1 e Mass fractions, elemental analysis, and equilibrium moisture contents (EMC) for wood specimens. Specimen Mass fraction (<100 mm)
Elemental analysis (mass%) C
H
O
72 h EMC (%)
Air Dried
0.08
46.3
6.2
47.5
21.6
220/5 220/30 220/60
0.26 0.53 0.76
46.4 47.5 48.2
5.9 6.0 6.0
47.7 46.5 45.8
11.1 9.4 7.6
260/5 260/30 260/60
0.46 0.70 0.89
49.3 51.1 50.5
5.7 5.7 5.8
45.0 43.2 43.6
8.2 5.8 8.2
300/5 300/30 300/60
0.79 0.87 0.88
49.4 54.2 56.7
6.0 5.7 5.5
44.7 40.1 37.8
9.2 22.4 28.0
95
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
The largest effect was seen with increased time of torrefaction rather than increasing temperature. Torrefaction at 260 C for 60 min and 300 C for either 30 or 60 min resulted in nearly 90% of the material having a particle size of 100 mm or less. Torrefaction at 220 C for 60 min showed increased grindability with respect to 260 C for 30 min. Results from torrefaction at 260 C and 300 C indicate that a plateau in grindability was reached, with optimal grindability achieved at either mild torrefaction conditions for at least 1 h or severe conditions for at least 30 min is appropriate. Elemental analysis showed a decrease in the %H and %O mass with increasing torrefaction severity (Table 1). This trend has been seen during wood coalification and results from a loss of methoxyl and hydroxyl groups on the aromatics structures of lignin [20] and degradation of the hemicelluloses and cellulose [21,22]. During torrefaction, changes occurred to the hydrophobicity of the wood chips with decreases in equilibrium moisture content (EMC) of specimens from 21.6% EMC in air dried wood to 11.1%, 9.4%, and 7.5% EMC for wood torrefied at 220 C for 5, 30, and 60 min respectively. At 260 C there was a further decrease in EMC from 8.2% at 5 min to 5.8% at 30 min. However, after torrefaction for 60 min the EMC was found to be 8.2%. This trend to higher EMC continued for torrefaction temperatures of 300 C with EMCs of 9.2%, 22.4%, and 28.0% for times of 5, 30, 60 min respectively (Table 1). In context of energy gains, it can be seen that at a torrefaction temperature of 300 C, any gain in EMC resulting from torrefaction is lost at torrefaction times of 30 or 60 min (Fig. 2). 13 C solid state NMR was used to characterize chemical changes that occurred during torrefaction. Some peaks and regions of the NMR spectra for air dried wood are presented as Fig. 3. In this spectrum carboxyl groups are mainly those present as acetyl groups on hemicelluloses, which is the main source
Fig. 2 e Equilibrium moisture content (EMC) as a function of energy gain before and after torrefaction.
Fig. 3 e Assignment of dried P. radiata.
13
C solid state NMR spectrum of air
of the acetate peak. The aromatic region and the methoxyl peaks are from lignin units. Peaks labelled ‘C’ are signals arising from cellulose, with C1 being the anomeric carbon and C6 being the methylol moiety [23]. Both C4 and C6 of cellulose exhibit two different chemical shifts, these arise from crystal interior (labelled ‘i’) or crystal surface (labelled ‘s’) chains of cellulose [24]. Chemical shift ranges were chosen to monitor chemical changes in the wood after torrefaction; these were the alkyl (0e50 ppm), O-alkyl (60e96 ppm), and carbonyl (185e220 ppm) regions based on assignments by Preston [25]. The alkyl region is indicative of the development of carbon side chain groups, the O-alkyl of the intact carbohydrates present, and the carbonyl region of the formation of aldehydes and ketones. Fig. 4 presents NMR spectra for wood torrefied at 220 C, 260 C, and 300 C for 60 min. As the temperature increased there is evidence for the degradation of the cellulose crystallites as shown by the decrease in intensity of the C4i peak relative to the C4s peak. This would imply that the surface area to volume ratio of the cellulose crystallites is changing in favour of more surface cellulose chains (i.e. smaller crystallites), although, due to the underlying peaks from hemicelluloses and lignin peaks under the C4s peak quantification is not possible. However, by taking the area of the C4i (89.3 ppm) peak and multiplying this by 6 (i.e. 6 carbon atoms per glucose molecule), and representing this value as a proportion of the total carbon area (0e220 ppm), it is possible to obtain an estimate of the proportion of crystal interior cellulose chains with respect to the total carbon. Air dried wood showed 25% crystal interior cellulose chains and after 60 min torrefaction, 220 C showed 20%, 260 C showed 16%, and 300 C showed 4%. The decrease in crystal interior
96
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
Fig. 4 e 13C solid state NMR spectra of wood torrefied at 220 C (c), 260 C (b), and 300 C (a).
chains is attributed to fragmentation of cellulose crystals and thermo-chemical degradation. In the carbonyl region there is little evidence for the formation of aldehydes and all signals in this region can be assigned to ketones, this absence of aldehydes has been observed previously by McGrath et al. during cellulose charring at 300 C [26] and is likely due to the formation of cyclopentenones and cyclohexenones [22]. Considering only the carbonyl, O-alkyl, and alkyl regions, it can be shown at greater torrefaction temperatures there is an increase in the appearance of both carbonyl and alkyl signals and a decrease in the O-alkyl region signals (Fig. 5). This is in accordance with decomposition of hemicelluloses occurring at temperatures above 220 C and decomposition of cellulose beginning around 300 C [4]. The increase in signal intensities in the alkyl and carbonyl regions can be attributed to rearrangement products from the breakdown of the hemicelluloses and cellulose into compounds such as furfurals [21] that then can react further to give a host of alkyl and aromatic compounds [22]. Cleavage of the lignin structure was observed with de-etherification of lignin guaiacyl ether units (G4e, 153 ppm) to free phenolics (G4f, 146 ppm) (Fig. 6). This de-etherification onsets at 220 C after 60 min, increases with time at 260 C, and plateaus at a torrefaction temperature of 300 C at approximately half the initial value seen in air dried wood. Synchrotron based X-ray yields high resolution diffractograms that overcome many of the short comings when using conventional X-ray sources in the study of wood [27]. Fig. 7 presents the diffractograms obtained from air dried wood and wood torrefied at 250 C and 275 C for 45e50 min.
Fig. 5 e Compositional change (%) of alkyl, O-alkyl, and carbonyl regions of 13C soils state NMR spectra of wood torrefied at 220 C (light hatching), 260 C (medium hatching), and 300 C (dense hatching) for 60 min.
Fig. 6 e Changes to the proportion of lignin guaiacyl etherified units to guaiacyl free phenolic units as determined by 13C solid state NMR.
97
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
Table 3 e Apparent cross-sectional unit cell parameters of wood specimens based on the cellulose Ib monoclinic crystal lattice. Specimen UM 250 C 275 C
Fig. 7 e X-ray diffractograms of air dried wood (line), wood torrefied at 250 C for 45 min (dot-dash), and 275 C for 50 min (dash).
A decrease in the 2q value for the d(20 0) peak is similar to that observed during traditional wood drying and indicates an increase in the spacing between the hydrogen-bonded cellulose crystal sheets [28,29]. Collapse of d(110) and d(10) has also been observed during wood drying and is indicative of changes in hydrogen-bonding in the crystal sheet plane [12]. No clear trend was observed in the cross-sectional apparent crystallite dimension under the three conditions examined, although, there is evidence for cellulose chain degradation at torrefaction conditions of 275 C. This is supported by a decrease in the average crystallite lateral dimension, the proportion of interior crystal chains and crystallinity index decreasing under these conditions (Table 2). As torrefaction temperature increases there is an increase in the distance between hydrogen-bonded sheets (a), and a decrease in both distance between chains in hydrogenbonded sheets (b) and monoclinic angle (g) (Table 3). This effect has been seen in traditional dried wood and has been suggested to be an effect of cellulose crystal dehydration and co-crystallisation of matrix components onto the cellulose crystal surfaces [30] or changes in the matrix volume
Table 2 e Calculated d-spacings, average lateral crystal dimensions, proportion of cellulose interior chains, and crystallinity indices of wood specimens. Specimen UM 250 C 275 C
dð110Þ (nm)
d(1 1 0) (nm)
d(2 0 0) (nm)
Laverage (nm)
X
C.I.
0.599 0.592 0.585
0.530 0.531 0.534
0.397 0.403 0.406
2.46 2.57 2.15
0.29 0.31 0.22
0.862 0.850 0.698
a (nm)
b (nm)
g ( )
0.800 0.810 0.815
0.800 0.780 0.772
97.0 96.2 95.2
either increasing or decreasing forces applied on the cellulose crystallites causing conformational changes [28]. The differential response with respect to unit directions a and b from the cellulose crystallites is likely due to the crystals mechanical anisotropy as the moduli of elasticity parallel and perpendicular to the fibre repeat axis (c) differ by an order of magnitude [31]. It has been proposed that during heating, the molecules on the cellulose crystallite surfaces undergo a conformational change from an inter-molecular hydrogen-bonding gaucheegauche and/or gauche-trans methylol groups to an intramolecular hydrogen-bonding trans-gauche conformation [12]. These conformational changes are likely to have started occurring during the temperature ramp to torrefaction temperature. This would have the effect of presenting a more hydrophobic cellulose crystal surface to the surrounding environment, decreasing the amount of moisture that could be adsorbed. The increase in hydrophobicity for light and mild torrefaction conditions was attributed to dehydration of cell-wall polymers, causing an increase in polymerepolymer interactions replacing polymerewater interactions. The polymerepolymer interactions are only partly reversible on adsorption of water. Under severe torrefaction conditions the polymers become chemically degraded, so the amount of adsorbed water reflects the hydrophilicity of the degradation products.
4.
Conclusions
The observed changes in moisture uptake were attributed to changes on the surfaces of cellulose crystallites. At lower torrefaction temperatures there occurred a modification of the molecular conformation at crystallite surfaces, creating relatively hydrophobic surfaces with little effect on the crystallite lateral dimensions or overall crystallinity. At higher torrefaction temperatures, thermal degradation caused a decrease in lateral crystal dimensions and crystallinity. The new, relatively hydrophilic surfaces accounted for increases in EMC. Although increased energy density and grindability are improved by increasing torrefaction temperature of P. radiata, moisture stability has an optimal processing temperature. Over the temperature-time ranges used the optimised conditions were 300 C for 5 min. At this temperature-time regime there was a 22.7% increase in energy density, a 71.5% increase in particles ground to less than 100 mm, and a decrease of 12.4% in EMC relative to air dried wood.
98
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
Acknowledgements The authors would like to thank Dr Nigel Kirby of the Australian Synchrotron SAXS/WAXS beam-line (Melbourne, Australia) for his time and kind assistance, and Dr Roger Newman (Scion) for his insightful comments. This study was supported through Scion’s Core Funding.
references
[1] Bourgois J, Guyonnet R. Characterization and analysis of torrefied wood. Wood Sci Technol 1988;22:143e55. [2] Bergman PCA, Boersma AR, Kiel JHA, Prins MJ, Ptasinski KJ, Janssen FJJG. Torrefaction for entrained-flow gasification of biomass. Report No.: ECN-Ce05067. The Netherlands: Energy Research Centre of the Netherlands (ECN); 200550. [3] Tjeerdsma BF, Boonstra M, Pizzi A, Tekely P, Militz H. Characterisation of thermally modified wood: molecular reason for wood performance improvement. Holz Roh Werkst 1998;56:149e53. [4] Prins MJ, Ptasinski KJ, Janssen JJG. Torrefaction of wood part 2. Analysis of products. J Anal Appl Pyrol 2006;77:35e40. [5] Arias B, Pevida C, Fermoso J, Plaza MG, Rubiera F, Pis JJ. Influence of torrefaction on the grindability and reactivity of woody biomass. Fuel Process Technol 2008;89:169e75. [6] van der Stelt MJC, Gerhauser H, Kiel JHA, Ptasinski KJ. Biomass upgrading by torrefaction for the production of biofuels: a review. Biomass Bioenerg 2011;9:3748e62. [7] Melkior T, Jacob S, Gerbaud G, Hediger S, Le Pape L, Bonnefois L, et al. NMR analysis of the transformation of wood constituents by torrefaction. Fuel 2012;92:271e80. [8] Chen W-H, Kuo P- C. Torrefaction and co-torrefaction characterization of hemicellulose, cellulose and lignin as well as torrefaction of some basic constituents in biomass. Energy 2011;36:803e11. [9] Spiller D. Effect of heat treatment on equilibrium moisture content of three New Zealand grown woods. New Zeal J Sci Technol 1948;30B:24e6. [10] Kininmonth JA. Effect of timber drying temperature on subsequent moisture and dimensional changes. New Zeal J For Sci 1976;6:101e7. [11] Ishimaru Y, Arai K, Mizutani M, Oshima K, Iida I. Physical and mechanical properties of wood after moisture conditioning. J Wood Sci 2001;47:185e91. [12] Hill SJ, Kirby NM, Mudie ST, Hawley AM, Ingham B, Franich RA, et al. Effect of drying and rewetting of wood on cellulose molecular packing. Holzforschung 2010;64:421e7. [13] Newman RH, Hemmingson JA. Carbon-13 nuclear magnetic resonance studies of kraft pulping. Holzforschung 1993;47:234e8. [14] Gjonnes J, Norman N, Viervoll H. The state of order in cellulose as revealed from X-ray diffractograms. Acta Chem Scand 1958;12:489e94.
[15] Davidson TC, Newman RH, Ryan MJ. Variations in the fibre repeat between samples of cellulose I from different sources. Carbohydr Res 2004;339:2889e93. [16] Sugiyama J, Vuong R, Chanzy H. Electron diffraction study on the two crystalline phases occurring in native cellulose from algal cell wall. Macromolecules 1991;24:4168e75. [17] Murdock C. The form of the x-ray diffraction bands for regular crystals of colloidal size. Phys Rev 1930;35:8e23. [18] Segal L, Creely JJ, Martin AE, Conrad CM. An empirical method for estimating the degree of crystallinity of native cellulose using the X-ray diffractometer. Text Res J 1959;29:786e94. [19] Prins MJ, Ptasinski KJ, Janssen JJG. Torrefaction of wood part 1. Weight loss kinetics. J Anal Appl Pyrol 2006;77:28e34. [20] Hatcher PG. Dipolar-dephasing 13C NMR studies of decomposed wood and coalified xylem tissue: evidence for chemical structural changes associated with defunctionalization of lignin structural units during coalification. Energ Fuel 1988;2:48e58. [21] Sanchez R, Rieumont J, Tavares MIB. Structural characterization of the polymeric lattice of furfuraldehyde resins. Polym Test 1998;17:395e401. [22] Pastorova I, Botto RE, Arisz PW, Boon JJ. Cellulose char structure: a combined analytical PyeGCeMS, FTIR, and NMR study. Carbohydr Res 1994;262:27e47. [23] Gil AM, Neto CP. Solid-state NMR studies of wood and other lignocellulosic materials. Ann R NMR S 1999;37:75e117. [24] Newman RH. Estimation of the lateral dimensions of cellulose crystallites using 13C NMR signal strengths. Solid State Nucl Mag 1999;15:21e9. [25] Preston CM, Sayer BG. What’s in a nutshell: an investigation of the structure by carbon-13 cross-polarization magic-angle spinning nuclear magnetic resonance spectroscopy. J Agr Food Chem 1992;40:206e10. [26] McGrath TE, Chan GW, Hajaligol MR. Low temperature mechanism for the formation of polycyclic aromatic hydrocarbons from the pyrolysis of cellulose. J Anal Appl Pyrol 2003;66:51e70. [27] Leppanen K, Andersson S, Torkkeli M, Knaapila M, Kotelnikova N, Serimaa R. Structures of cellulose and microcrystalline cellulose from various wood species, cotton, and flax studied by X-ray scattering. Cellulose 2009;16:999e1015. [28] Abe K, Yamamoto H. Mechanical interaction between cellulose microfibril and matrix substance in wood cell wall determination by X-ray diffraction. J Wood Sci 2005;51:334e8. [29] Sugino H, Sugimoto H, Miki T, Kanayama K. Fine structure changes of wood during moisture adsorption and desorption process analyzed by X-ray diffraction measurement. Mokuzai Gakkaishi 2007;53:82e9. [30] Newman RH, Hemmingson JA. Interactions between locust bean gum and cellulose characterized by 13C n.m.r. spectroscopy. Carbohydr Polym 1998;36:167e72. [31] Diddens I, Murphy B, Krisch M, Muller M. Anisotropic elastic properties of cellulose measured using inelastic X-ray scattering. Macromolecules 2008;41:9755e9.
Available online at www.sciencedirect.com
http://www.elsevier.com/locate/biombioe
Chemical and cellulose crystallite changes in Pinus radiata during torrefaction Stefan J. Hill*, Warren J. Grigsby, Peter W. Hall Scion, Private Bag 3020, Rotorua 3046, New Zealand
article info
abstract
Article history:
The impact on the chemical composition and changes to the cellulose crystallites in Pinus
Received 1 July 2012
radiata wood chips under light (ca. 230 C), mild (ca. 260 C), and severe (290þ C) torrefaction
Received in revised form
temperatures at a range of times was examined by solid state
25 April 2013
Dephased NMR spectroscopy, TGA, and synchrotron based X-ray diffraction. Results
Accepted 26 April 2013
indicated the decomposition of hemicelluloses into furfurals at the lowest temperature
Available online 30 May 2013
with little modification to the lignin or cellulose. De-polymerisation of lignin and cellulose
13
C CP-MAS and Dipolar
was observed as torrefaction severity increased. The increased hydrophobicity under light Keywords:
and mild torrefaction severity was attributed to a combination of thermo-chemical mod-
Moisture content
ifications to hemicelluloses and lignin, along with cellulose crystal lattice changes. The
Torrefaction
observed decrease in hydrophobicity under severe torrefaction conditions was attributed
Softwood
to the degradation of cellulose crystallites.
Solid state NMR
ª 2013 Elsevier Ltd. All rights reserved.
X-ray diffraction
1.
Introduction
Torrefaction is the semi-conversion of wood to charcoal by pyrolysing lignocellulosics at temperatures ranging from 200 to 300 C in an inert or semi-inert atmosphere at standard pressures [1]. This has the effect of improving not only the energy density [2], but also moisture stability [3,4] and grindability [5] of the woody biomass. This is beneficial when considering transport costs, long term storage of torrefied wood [6], and processing into pulverised fuel. The torrefaction process has been shown to cause many changes to the chemical composition of woody biomass depending on the temperature and duration of heating [7]. These changes include modifications to the lignin and hemicelluloses at lower torrefaction temperature and degradation of cellulose, lignin, and the hemicelluloses at higher torrefaction temperatures.
Previous studies have illuminated some of the chemical processes occurring during torrefaction utilising techniques such as NMR, FT-IR, and X-ray diffraction to name a few [1,3]. Chen and Kuo [8] defined three temperature regimes of torrefaction; light torrefaction at temperatures ca. 230 C, mild torrefaction at temperatures ca. 260 C, and severe torrefaction at temperatures of 290 C and above. These definitions can be described in terms of the thermal modification extent of the wood components during torrefaction. Under light conditions the mass fractionloss of the cellulose (1.1%), hemicelluloses (2.7%), and lignin (1.5%) is minimal. Under mild conditions and the mass fraction losses increase to 4.4%, 38.0%, and 3.1% respectively, meaning that at ca. 260 C there is proportionally significantly more degradation of the hemicelluloses than either cellulose or lignin. Under the severe torrefaction regime further mass fraction losses in the three components occur: 44.8% for cellulose, 58.3% for the hemicelluloses, and 7.0% for
* Corresponding author. Scion, 49 Sala St, Rotorua 3010, New Zealand. Tel.: þ64 7 343 5872; fax: þ64 7 343 5507. E-mail address: [email protected] (S.J. Hill). 0961-9534/$ e see front matter ª 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biombioe.2013.04.025
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
lignin. It should be noted that these substances were tested as pure compounds and not in wood [8]. These provide a useful guide to the influence of temperature on the thermal degradation of wood. It has been long known that the thermal modification of wood leads to an increase in hydrophobicity and that this is a consequence of changes to the nano-structure of the wood [9,10]. This observation has been attributed to coupling and decoupling of hydroxyl groups on the cellulose crystal surface [11]. The nature of these changes has more recently been associated in part to partially irreversible changes to the cellulose chain conformations on crystal surfaces [12]. In the current study two technical properties of Pinus radiata wood, notably hydrophobicity and grindability, were studied as potential functions of overall chemical and nanostructural changes. This was achieved by torrefying wood at several combinations of temperature and times. Specimens were taken for characterisation by NMR, X-ray diffraction, TGA, and elemental analysis.
2.
Materials & methods
2.1.
Sampling
The trees (Pinus radiata D. Don) used in this study were w28 years old, sourced from log production site 191100E 5742800N (NZTM) in Kaingaroa forest in May of 2010, on the volcanic plateau in the Central North Island of New Zealand (38 340 S, 176 320 E). The wood chip used in the experiments was recovered from chipped sawmilling slab wood, and so is from the outer more dense part of the stem, with the log being cut from the lower part (bottom 12 m) of the tree. Due to the industrial nature of the chipping process, the sample should be regarded as a bulked up average. Logs were processed from stump to mill in 3e4 days, with open air storage at harvest and sawmill sites.
2.2.
Torrefaction
A torrefaction test rig used a small electric furnace fitted with a torrefaction reactor tube diameter (54 mm diameter by 240 mm long; 0.55 L). The reactor tube was fed with argon gas which was preheated in a copper coil before entering the reactor tube. The temperature of the furnace, the internal temperature of the reactor tube (sides and centre), and the electrical power use of the furnace were monitored during torrefaction. Moisture contents (MC%) of samples were calculated as a percentage of asamples 105 C oven-dried weight: MCð%Þ ¼ ðOriginal massðgÞ=Oven dried massðgÞ 1Þ 100 (1) Torrefaction experiments used air dried (average moisture content 18%) P. radiata chips (9e26 mm in size), temperatures of 220 C, 260 C, and 300 C, and torrefaction times of 5, 30, and 60 min. In all cases the heating rate of the furnace was 8 C$min1 until the target temperature was reached. The
93
specimens were then held at the target temperature for the duration of the torrefaction experiment. After the prescribed time the rig was opened and the specimen was removed and allowed to air-cool. Sub-samples of these specimens were used for NMR, TGA, hydrophobicity, grindability, and elemental analysis experiments. A further two specimens were torrefied at 250 C and 275 C for 45e50 min, i.e. at temperatures identified as being in the range of particular interest, for use in synchrotron-sourced X-ray diffraction studies. These specimens were conditioned at 22 C and 32% relative humidity for at least 48 h prior to further analysis.
2.3.
NMR
The samples were spun at 5 kHz in a 4 mm Bruker SB magicangle spinning probe, for 13C NMR at 50.3 MHz using a Bruker 200 DRX spectrometer. For the standard cross-polarisation (CP) experiment, each 1.5 s pulse delay was followed by a proton preparation pulse of duration tp ¼ 4.6 ms, a 1 ms contact time and a 30 ms acquisition time. The proton transmitter power was increased to a value corresponding to a 90 pulse width of 2.8 ms for proton decoupled during 13C data acquisition. In case of dipolar dephasing experiments a dephasing delay of 70 ms was introduced on the carbon channel prior to decoupled data acquisition. Transients were averaged over 10k transients for CP and 60k transients for dipolar dephasing experiments. All spectra had a Gaussian line broadening of 25 Hz applied prior to Fourier transform and were calibrated so that the cellulose interior C4 peak was assigned a value of 89.3 ppm, previously established relative to polydimethylsilane at 1.96 ppm, in turn measured relative to tetramethylsilane at 0 ppm [13].
2.4.
X-ray diffraction
A synchrotron radiation sourced X-ray beam energy of 13 keV ˚ . A Mar-CCD was used, giving an X-ray wavelength of 0.9537 A detector (1024 1024 pixels) was located 120 mm from the specimen allowing detection of 2q angles out to approximately 35 . The beam size was 0.2 mm by 0.1 mm, and the detector pixel size was 0.158 mm2. The exposure time was 2 s with nine data points taken per specimen spaced 2e3 mm apart to obtain nine diffraction patterns per specimen. The diffractograms presented are the average of the nine data sets obtained per specimen. Diffraction traces were extracted from diffraction patterns using an approximate 10 cake slice orthogonal to the grain direction. From this an approximate 10 cake slice centred approximately 30 off-axis from the grain direction was subtracted. This subtraction removes the background from the diffuse halo of reflections of non-cellulose origin such as hemicelluloses and water [12]. A Lorentzian function was used to fit the major peaks assuming, for simplicity, a totally cellulose Ib structure (d(10), d(1 1 0), d(200)) [14]. Although P. radiata consists of a mixture of cellulose Ia and Ib crystal forms of almost equal proportions [15], with peaks that overlap significantly, the major structural difference seen is in the fibre repeat direction and as such is not significant when calculating cross-sectional dimensions [16].
94
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
The minimum lateral crystal dimension (L) was calculated using the Scherrer equation: Lhkl ¼ Kl=ßhkl cosqhkl
(2)
where K is a crystal geometry parameter, approximated as 0.93 [17], l is a x-ray wavelength, b is the width at half peak height, and q is the Bragg angle. As the value of L is dependent on the crystal plane chosen, Lð110Þ , L(1 1 0), and L(2 0 0) were calculated, and averaged to give an Laverage value. An estimate of the proportion of surface to interior cellulose crystal chains was made using [15]: 2 X ¼ Laverage 2h =L2average
(3)
where L is the calculated minimum lateral crystal dimension and h is the thickness of a cellulose chain (0.57 nm). An estimate of changes in crystallinity was determined by the empirical method of Segal [18]: C:I: ¼ Ið2 0 0Þ Iam =Ið2 0 0Þ
(4)
where I(2 0 0) is the height of the (2 0 0) peak, and Iam is the height of the minima between the d(1 1 0) and the d(2 0 0) peaks.
2.5.
TGA
Thermogravimetric analysis (TGA) was undertaken to simulate the torrefaction process using a Thermal Analysis Instruments Q500 instrument under a nitrogen atmosphere. Air dried wood specimens were heated at a rate of 50 C$min1 until the required isothermal temperature (220, 260, or 300 C) was reached. The specimens were held at this isothermal temperature for a further 50 min with weight loss recorded.
2.6.
Other analysis
Elemental analysis (%C/%H) was carried out by The Campbell Microanalytical Laboratory (University of Otago, New Zealand), %O was estimated by difference. Grindability was determined by grinding samples in a ring-mill for 1 min with the resulting material sieved to determine particle size distributions (>500 mm, 100e500 mm, and <100 mm).
3.
Results & discussion
Mass loss of P. radiata chips during torrefaction was measured from TGA experiments, with mass losses after 10 min of 7.3%, 14.8%, and 36.6% for 220 C, 260 C, and 300 C respectively. After 60 min this loss increased to 8.9%, 27.7%, and 68.2% for 220 C, 260 C, and 300 C respectively (Fig. 1). These weight loss figures are consistent with those previously reported for willow at similar temperatures and times where the mass loss was suggested to be a two-step kinetic process, with the first being the decomposition of hemicelluloses (Ea 76.0 kJ$mol1) and the second the decomposition of cellulose (Ea 151.7 kJ$mol1) [19]. A study by Chen and Kuo of the mass loss of individual lignocellulosic components (cellulose, hemicelluloses, and lignin) agrees with this two-step kinetic description, with lignin loss being only a relatively
Fig. 1 e TGA analysis of mass loss of wood at 220 C (line), 260 C (dash), and 300 C (dot-dash).
minor contributor to mass loss at these temperatures studied [8]. Mass loss during actual torrefaction was found to be 3.2%, 17.0%, and 42.6% for 220, 260, and 300 C respectively, all after 60 min. It can be seen in all cases that the TGA mass losses overestimate the amount of mass loss observed in the torrefaction system used. The difference is attributed to a combination of increased heating ramp and larger surface area of the small samples required for TGA analysis. Increased time and temperature of torrefaction was associated with an increase in grindability and resulted in greater fraction of sub-100 mm particle sizes (Table 1).
Table 1 e Mass fractions, elemental analysis, and equilibrium moisture contents (EMC) for wood specimens. Specimen Mass fraction (<100 mm)
Elemental analysis (mass%) C
H
O
72 h EMC (%)
Air Dried
0.08
46.3
6.2
47.5
21.6
220/5 220/30 220/60
0.26 0.53 0.76
46.4 47.5 48.2
5.9 6.0 6.0
47.7 46.5 45.8
11.1 9.4 7.6
260/5 260/30 260/60
0.46 0.70 0.89
49.3 51.1 50.5
5.7 5.7 5.8
45.0 43.2 43.6
8.2 5.8 8.2
300/5 300/30 300/60
0.79 0.87 0.88
49.4 54.2 56.7
6.0 5.7 5.5
44.7 40.1 37.8
9.2 22.4 28.0
95
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
The largest effect was seen with increased time of torrefaction rather than increasing temperature. Torrefaction at 260 C for 60 min and 300 C for either 30 or 60 min resulted in nearly 90% of the material having a particle size of 100 mm or less. Torrefaction at 220 C for 60 min showed increased grindability with respect to 260 C for 30 min. Results from torrefaction at 260 C and 300 C indicate that a plateau in grindability was reached, with optimal grindability achieved at either mild torrefaction conditions for at least 1 h or severe conditions for at least 30 min is appropriate. Elemental analysis showed a decrease in the %H and %O mass with increasing torrefaction severity (Table 1). This trend has been seen during wood coalification and results from a loss of methoxyl and hydroxyl groups on the aromatics structures of lignin [20] and degradation of the hemicelluloses and cellulose [21,22]. During torrefaction, changes occurred to the hydrophobicity of the wood chips with decreases in equilibrium moisture content (EMC) of specimens from 21.6% EMC in air dried wood to 11.1%, 9.4%, and 7.5% EMC for wood torrefied at 220 C for 5, 30, and 60 min respectively. At 260 C there was a further decrease in EMC from 8.2% at 5 min to 5.8% at 30 min. However, after torrefaction for 60 min the EMC was found to be 8.2%. This trend to higher EMC continued for torrefaction temperatures of 300 C with EMCs of 9.2%, 22.4%, and 28.0% for times of 5, 30, 60 min respectively (Table 1). In context of energy gains, it can be seen that at a torrefaction temperature of 300 C, any gain in EMC resulting from torrefaction is lost at torrefaction times of 30 or 60 min (Fig. 2). 13 C solid state NMR was used to characterize chemical changes that occurred during torrefaction. Some peaks and regions of the NMR spectra for air dried wood are presented as Fig. 3. In this spectrum carboxyl groups are mainly those present as acetyl groups on hemicelluloses, which is the main source
Fig. 2 e Equilibrium moisture content (EMC) as a function of energy gain before and after torrefaction.
Fig. 3 e Assignment of dried P. radiata.
13
C solid state NMR spectrum of air
of the acetate peak. The aromatic region and the methoxyl peaks are from lignin units. Peaks labelled ‘C’ are signals arising from cellulose, with C1 being the anomeric carbon and C6 being the methylol moiety [23]. Both C4 and C6 of cellulose exhibit two different chemical shifts, these arise from crystal interior (labelled ‘i’) or crystal surface (labelled ‘s’) chains of cellulose [24]. Chemical shift ranges were chosen to monitor chemical changes in the wood after torrefaction; these were the alkyl (0e50 ppm), O-alkyl (60e96 ppm), and carbonyl (185e220 ppm) regions based on assignments by Preston [25]. The alkyl region is indicative of the development of carbon side chain groups, the O-alkyl of the intact carbohydrates present, and the carbonyl region of the formation of aldehydes and ketones. Fig. 4 presents NMR spectra for wood torrefied at 220 C, 260 C, and 300 C for 60 min. As the temperature increased there is evidence for the degradation of the cellulose crystallites as shown by the decrease in intensity of the C4i peak relative to the C4s peak. This would imply that the surface area to volume ratio of the cellulose crystallites is changing in favour of more surface cellulose chains (i.e. smaller crystallites), although, due to the underlying peaks from hemicelluloses and lignin peaks under the C4s peak quantification is not possible. However, by taking the area of the C4i (89.3 ppm) peak and multiplying this by 6 (i.e. 6 carbon atoms per glucose molecule), and representing this value as a proportion of the total carbon area (0e220 ppm), it is possible to obtain an estimate of the proportion of crystal interior cellulose chains with respect to the total carbon. Air dried wood showed 25% crystal interior cellulose chains and after 60 min torrefaction, 220 C showed 20%, 260 C showed 16%, and 300 C showed 4%. The decrease in crystal interior
96
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
Fig. 4 e 13C solid state NMR spectra of wood torrefied at 220 C (c), 260 C (b), and 300 C (a).
chains is attributed to fragmentation of cellulose crystals and thermo-chemical degradation. In the carbonyl region there is little evidence for the formation of aldehydes and all signals in this region can be assigned to ketones, this absence of aldehydes has been observed previously by McGrath et al. during cellulose charring at 300 C [26] and is likely due to the formation of cyclopentenones and cyclohexenones [22]. Considering only the carbonyl, O-alkyl, and alkyl regions, it can be shown at greater torrefaction temperatures there is an increase in the appearance of both carbonyl and alkyl signals and a decrease in the O-alkyl region signals (Fig. 5). This is in accordance with decomposition of hemicelluloses occurring at temperatures above 220 C and decomposition of cellulose beginning around 300 C [4]. The increase in signal intensities in the alkyl and carbonyl regions can be attributed to rearrangement products from the breakdown of the hemicelluloses and cellulose into compounds such as furfurals [21] that then can react further to give a host of alkyl and aromatic compounds [22]. Cleavage of the lignin structure was observed with de-etherification of lignin guaiacyl ether units (G4e, 153 ppm) to free phenolics (G4f, 146 ppm) (Fig. 6). This de-etherification onsets at 220 C after 60 min, increases with time at 260 C, and plateaus at a torrefaction temperature of 300 C at approximately half the initial value seen in air dried wood. Synchrotron based X-ray yields high resolution diffractograms that overcome many of the short comings when using conventional X-ray sources in the study of wood [27]. Fig. 7 presents the diffractograms obtained from air dried wood and wood torrefied at 250 C and 275 C for 45e50 min.
Fig. 5 e Compositional change (%) of alkyl, O-alkyl, and carbonyl regions of 13C soils state NMR spectra of wood torrefied at 220 C (light hatching), 260 C (medium hatching), and 300 C (dense hatching) for 60 min.
Fig. 6 e Changes to the proportion of lignin guaiacyl etherified units to guaiacyl free phenolic units as determined by 13C solid state NMR.
97
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
Table 3 e Apparent cross-sectional unit cell parameters of wood specimens based on the cellulose Ib monoclinic crystal lattice. Specimen UM 250 C 275 C
Fig. 7 e X-ray diffractograms of air dried wood (line), wood torrefied at 250 C for 45 min (dot-dash), and 275 C for 50 min (dash).
A decrease in the 2q value for the d(20 0) peak is similar to that observed during traditional wood drying and indicates an increase in the spacing between the hydrogen-bonded cellulose crystal sheets [28,29]. Collapse of d(110) and d(10) has also been observed during wood drying and is indicative of changes in hydrogen-bonding in the crystal sheet plane [12]. No clear trend was observed in the cross-sectional apparent crystallite dimension under the three conditions examined, although, there is evidence for cellulose chain degradation at torrefaction conditions of 275 C. This is supported by a decrease in the average crystallite lateral dimension, the proportion of interior crystal chains and crystallinity index decreasing under these conditions (Table 2). As torrefaction temperature increases there is an increase in the distance between hydrogen-bonded sheets (a), and a decrease in both distance between chains in hydrogenbonded sheets (b) and monoclinic angle (g) (Table 3). This effect has been seen in traditional dried wood and has been suggested to be an effect of cellulose crystal dehydration and co-crystallisation of matrix components onto the cellulose crystal surfaces [30] or changes in the matrix volume
Table 2 e Calculated d-spacings, average lateral crystal dimensions, proportion of cellulose interior chains, and crystallinity indices of wood specimens. Specimen UM 250 C 275 C
dð110Þ (nm)
d(1 1 0) (nm)
d(2 0 0) (nm)
Laverage (nm)
X
C.I.
0.599 0.592 0.585
0.530 0.531 0.534
0.397 0.403 0.406
2.46 2.57 2.15
0.29 0.31 0.22
0.862 0.850 0.698
a (nm)
b (nm)
g ( )
0.800 0.810 0.815
0.800 0.780 0.772
97.0 96.2 95.2
either increasing or decreasing forces applied on the cellulose crystallites causing conformational changes [28]. The differential response with respect to unit directions a and b from the cellulose crystallites is likely due to the crystals mechanical anisotropy as the moduli of elasticity parallel and perpendicular to the fibre repeat axis (c) differ by an order of magnitude [31]. It has been proposed that during heating, the molecules on the cellulose crystallite surfaces undergo a conformational change from an inter-molecular hydrogen-bonding gaucheegauche and/or gauche-trans methylol groups to an intramolecular hydrogen-bonding trans-gauche conformation [12]. These conformational changes are likely to have started occurring during the temperature ramp to torrefaction temperature. This would have the effect of presenting a more hydrophobic cellulose crystal surface to the surrounding environment, decreasing the amount of moisture that could be adsorbed. The increase in hydrophobicity for light and mild torrefaction conditions was attributed to dehydration of cell-wall polymers, causing an increase in polymerepolymer interactions replacing polymerewater interactions. The polymerepolymer interactions are only partly reversible on adsorption of water. Under severe torrefaction conditions the polymers become chemically degraded, so the amount of adsorbed water reflects the hydrophilicity of the degradation products.
4.
Conclusions
The observed changes in moisture uptake were attributed to changes on the surfaces of cellulose crystallites. At lower torrefaction temperatures there occurred a modification of the molecular conformation at crystallite surfaces, creating relatively hydrophobic surfaces with little effect on the crystallite lateral dimensions or overall crystallinity. At higher torrefaction temperatures, thermal degradation caused a decrease in lateral crystal dimensions and crystallinity. The new, relatively hydrophilic surfaces accounted for increases in EMC. Although increased energy density and grindability are improved by increasing torrefaction temperature of P. radiata, moisture stability has an optimal processing temperature. Over the temperature-time ranges used the optimised conditions were 300 C for 5 min. At this temperature-time regime there was a 22.7% increase in energy density, a 71.5% increase in particles ground to less than 100 mm, and a decrease of 12.4% in EMC relative to air dried wood.
98
b i o m a s s a n d b i o e n e r g y 5 6 ( 2 0 1 3 ) 9 2 e9 8
Acknowledgements The authors would like to thank Dr Nigel Kirby of the Australian Synchrotron SAXS/WAXS beam-line (Melbourne, Australia) for his time and kind assistance, and Dr Roger Newman (Scion) for his insightful comments. This study was supported through Scion’s Core Funding.
references
[1] Bourgois J, Guyonnet R. Characterization and analysis of torrefied wood. Wood Sci Technol 1988;22:143e55. [2] Bergman PCA, Boersma AR, Kiel JHA, Prins MJ, Ptasinski KJ, Janssen FJJG. Torrefaction for entrained-flow gasification of biomass. Report No.: ECN-Ce05067. The Netherlands: Energy Research Centre of the Netherlands (ECN); 200550. [3] Tjeerdsma BF, Boonstra M, Pizzi A, Tekely P, Militz H. Characterisation of thermally modified wood: molecular reason for wood performance improvement. Holz Roh Werkst 1998;56:149e53. [4] Prins MJ, Ptasinski KJ, Janssen JJG. Torrefaction of wood part 2. Analysis of products. J Anal Appl Pyrol 2006;77:35e40. [5] Arias B, Pevida C, Fermoso J, Plaza MG, Rubiera F, Pis JJ. Influence of torrefaction on the grindability and reactivity of woody biomass. Fuel Process Technol 2008;89:169e75. [6] van der Stelt MJC, Gerhauser H, Kiel JHA, Ptasinski KJ. Biomass upgrading by torrefaction for the production of biofuels: a review. Biomass Bioenerg 2011;9:3748e62. [7] Melkior T, Jacob S, Gerbaud G, Hediger S, Le Pape L, Bonnefois L, et al. NMR analysis of the transformation of wood constituents by torrefaction. Fuel 2012;92:271e80. [8] Chen W-H, Kuo P- C. Torrefaction and co-torrefaction characterization of hemicellulose, cellulose and lignin as well as torrefaction of some basic constituents in biomass. Energy 2011;36:803e11. [9] Spiller D. Effect of heat treatment on equilibrium moisture content of three New Zealand grown woods. New Zeal J Sci Technol 1948;30B:24e6. [10] Kininmonth JA. Effect of timber drying temperature on subsequent moisture and dimensional changes. New Zeal J For Sci 1976;6:101e7. [11] Ishimaru Y, Arai K, Mizutani M, Oshima K, Iida I. Physical and mechanical properties of wood after moisture conditioning. J Wood Sci 2001;47:185e91. [12] Hill SJ, Kirby NM, Mudie ST, Hawley AM, Ingham B, Franich RA, et al. Effect of drying and rewetting of wood on cellulose molecular packing. Holzforschung 2010;64:421e7. [13] Newman RH, Hemmingson JA. Carbon-13 nuclear magnetic resonance studies of kraft pulping. Holzforschung 1993;47:234e8. [14] Gjonnes J, Norman N, Viervoll H. The state of order in cellulose as revealed from X-ray diffractograms. Acta Chem Scand 1958;12:489e94.
[15] Davidson TC, Newman RH, Ryan MJ. Variations in the fibre repeat between samples of cellulose I from different sources. Carbohydr Res 2004;339:2889e93. [16] Sugiyama J, Vuong R, Chanzy H. Electron diffraction study on the two crystalline phases occurring in native cellulose from algal cell wall. Macromolecules 1991;24:4168e75. [17] Murdock C. The form of the x-ray diffraction bands for regular crystals of colloidal size. Phys Rev 1930;35:8e23. [18] Segal L, Creely JJ, Martin AE, Conrad CM. An empirical method for estimating the degree of crystallinity of native cellulose using the X-ray diffractometer. Text Res J 1959;29:786e94. [19] Prins MJ, Ptasinski KJ, Janssen JJG. Torrefaction of wood part 1. Weight loss kinetics. J Anal Appl Pyrol 2006;77:28e34. [20] Hatcher PG. Dipolar-dephasing 13C NMR studies of decomposed wood and coalified xylem tissue: evidence for chemical structural changes associated with defunctionalization of lignin structural units during coalification. Energ Fuel 1988;2:48e58. [21] Sanchez R, Rieumont J, Tavares MIB. Structural characterization of the polymeric lattice of furfuraldehyde resins. Polym Test 1998;17:395e401. [22] Pastorova I, Botto RE, Arisz PW, Boon JJ. Cellulose char structure: a combined analytical PyeGCeMS, FTIR, and NMR study. Carbohydr Res 1994;262:27e47. [23] Gil AM, Neto CP. Solid-state NMR studies of wood and other lignocellulosic materials. Ann R NMR S 1999;37:75e117. [24] Newman RH. Estimation of the lateral dimensions of cellulose crystallites using 13C NMR signal strengths. Solid State Nucl Mag 1999;15:21e9. [25] Preston CM, Sayer BG. What’s in a nutshell: an investigation of the structure by carbon-13 cross-polarization magic-angle spinning nuclear magnetic resonance spectroscopy. J Agr Food Chem 1992;40:206e10. [26] McGrath TE, Chan GW, Hajaligol MR. Low temperature mechanism for the formation of polycyclic aromatic hydrocarbons from the pyrolysis of cellulose. J Anal Appl Pyrol 2003;66:51e70. [27] Leppanen K, Andersson S, Torkkeli M, Knaapila M, Kotelnikova N, Serimaa R. Structures of cellulose and microcrystalline cellulose from various wood species, cotton, and flax studied by X-ray scattering. Cellulose 2009;16:999e1015. [28] Abe K, Yamamoto H. Mechanical interaction between cellulose microfibril and matrix substance in wood cell wall determination by X-ray diffraction. J Wood Sci 2005;51:334e8. [29] Sugino H, Sugimoto H, Miki T, Kanayama K. Fine structure changes of wood during moisture adsorption and desorption process analyzed by X-ray diffraction measurement. Mokuzai Gakkaishi 2007;53:82e9. [30] Newman RH, Hemmingson JA. Interactions between locust bean gum and cellulose characterized by 13C n.m.r. spectroscopy. Carbohydr Polym 1998;36:167e72. [31] Diddens I, Murphy B, Krisch M, Muller M. Anisotropic elastic properties of cellulose measured using inelastic X-ray scattering. Macromolecules 2008;41:9755e9.